How to show that higher derivative theories (mostly) breaks unitarity?
Spinor field $\psi_{a_{1}...a_{n}\dot {b}_{1}..\dot {b}_{m}} $, which refer to the $\left( \frac{n}{2}, \frac{m}{2} \right)$ representation of the Lorentz group, must satisfy
$$ (\partial^{2} + M^{2})\psi_{a_{1}...a_{n}\dot {b}_{1}..\dot {b}_{m}} = 0 , \quad \partial^{\dot {b}_{1}a_{1}}\psi_{a_{1}...a_{n}\dot {b}_{1}..\dot {b}_{m}} = 0 , $$ for representing the unitary one-particle state of the Poincare group with spin $n + m$, which can be integer or half-integer, and mass $M$.
Then, these conditions can be combined in some field equation, which refers to the some Lagrangian. There is a requirement that the equation must not contain derivatives higher than second-order. If the requirement is violated, one says that it breaks the unitarity of the theory of corresponding field.
So, the question: how exactly can be showed that it breaks the unitarity? And does it always break the unitarity?